Model Explainer

Feature Importances

Feature Importances

Model performance metrics

metric Score
accuracy 0.78
precision 0.417
recall 0.346
f1 0.379
roc_auc_score 0.677
pr_auc_score 0.355
log_loss 1.14

Confusion Matrix

How many false positives and false negatives?

Precision Plot

Does fraction positive increase with predicted probability?

Classification Plot

Distribution of labels above and below cutoff

ROC AUC Plot

Trade-off between False positives and false negatives

PR AUC Plot

Trade-off between Precision and Recall

Lift Curve

Performance how much better than random?

Cumulative Precision

Expected distribution for highest scores

Individual Predictions

Select Random Index

Selected index: 2422

Prediction

label probability
0* 12.8 %
1 87.2 %

Contributions Plot

How has each feature contributed to the prediction?

Partial Dependence Plot

Contributions Table

How has each feature contributed to the prediction?
Reason Effect
Average of population 23.78%
crime_type_Theft_1 = 1.0 +28.44%
crime_month_4_0 = 0.0 +22.59%
crime_type_Violence and sexual offences_0 = 1.0 +7.13%
crime_year_2019_1 = 1.0 +5.59%
crime_type_Drugs_1 = 0.0 -3.43%
crime_type_Criminal damage and arson_1 = 0.0 +2.06%
LSOA_code = 0.22689075767993927 -1.96%
crime_month_11_0 = 1.0 +1.85%
crime_month_7_1 = 0.0 -1.68%
crime_year_2021_1 = 0.0 +1.47%
crime_type_Vehicle crime_0 = 1.0 -1.4%
crime_month_8_1 = 0.0 +0.85%
crime_month_9_0 = 1.0 +0.74%
crime_type_Public order_0 = 1.0 +0.67%
crime_month_1_1 = 0.0 +0.51%
crime_type_Possession of weapons_0 = 1.0 +0.0%
crime_year_2020_0 = 1.0 +0.0%
crime_type_Robbery_1 = 0.0 +0.0%
crime_month_3_0 = 1.0 +0.0%
crime_month_5_1 = 0.0 +0.0%
crime_month_10_0 = 1.0 +0.0%
crime_month_2_1 = 0.0 +0.0%
crime_month_6_0 = 1.0 +0.0%
crime_type_Burglary_1 = 0.0 +0.0%
Other features combined +0.0%
Final prediction 87.23%

What if...

Select Random Index

Selected index: 61080

Prediction

label probability
0 99.7 %
1 0.3 %

Feature Input

Adjust the feature values to change the prediction

Contributions Plot

How has each feature contributed to the prediction?

Partial Dependence Plot

Contributions Table

How has each feature contributed to the prediction?
Reason Effect
Average of population 23.78%
crime_type_Drugs_1 = 0.0 -5.66%
crime_type_Theft_1 = 0.0 -4.85%
LSOA_code = 0.1805555522441864 -4.36%
crime_type_Vehicle crime_0 = 1.0 -1.72%
crime_type_Violence and sexual offences_0 = 0.0 -2.31%
crime_month_7_1 = 0.0 -0.6%
crime_month_4_0 = 1.0 -0.33%
crime_type_Burglary_1 = 0.0 +0.0%
crime_year_2019_1 = 0.0 -0.5%
crime_type_Possession of weapons_0 = 1.0 +0.0%
crime_type_Criminal damage and arson_1 = 0.0 +0.0%
crime_month_11_0 = 1.0 +0.0%
crime_type_Public order_0 = 1.0 +0.0%
crime_year_2021_1 = 0.0 +0.0%
crime_month_1_1 = 0.0 +0.0%
crime_month_9_0 = 0.0 -2.97%
crime_month_3_0 = 1.0 -0.21%
crime_month_5_1 = 0.0 +0.0%
crime_month_8_1 = 0.0 +0.0%
crime_month_6_0 = 1.0 +0.0%
crime_type_Robbery_1 = 0.0 +0.0%
crime_month_2_1 = 0.0 +0.0%
crime_month_10_0 = 1.0 +0.0%
crime_year_2020_0 = 0.0 +0.0%
Other features combined +0.0%
Final prediction 0.27%

Feature Dependence

Shap Summary

Ordering features by shap value

Shap Dependence

Relationship between feature value and SHAP value